I found a new type of mechanical switch that might work well for the diode experiments, a Knife switch –>
This particular switch is sold by Radio Shack, but is also sold by a store in UK. Please see the Diode Lab homepage for details where to buy this switch in the UK and USA, and how it would be used for the diode experiments. Although I have not tested this switch yet, it should have ultra high off resistance far above 10 Tohm.
[continued from New diode testing design]
Each latched reed relays will be powered by a battery that is controlled by a phototransistor, which is controlled by a beam of light shining through a pin hole in the hammond shield. The phototransistor will be inside the Hammond shield. The LED that shines light on the phototransistor will be outside of the Hammond shield. The external LED will be connected to an external switch and battery. So momentarily pressing the switch, the latched reed relay will toggle.
The present design for the new diode testing setup is same as previously outlined with one exception. There will be *NO* electrical wires entering or exiting the Hammond shield. I will go back to my original fiber optic cable method.
A fiber optic cable consists of the fiber, which is usually coated with a resin buffer layer, which is surround by a jacket layer, usually plastic or rubber. Roughly one inch in length of the thick rubber from the end of the optic cable will be removed. So one inch of the end of the optic cable will be thin, and will go inside the side of the Hammond chassis. The end of the optic cable inside the chassis will be mounted to an LED, which will be driven by an LED driver, which will be connected to the electrometer.
Here’s a skematic diagram of the electrometer ->
That is it! I have tested this electrometer in various configurations, included the addition of the 100 Mohm resistors on the input pins (Vin+, Vin-). The rate at which this op-amp floats is very slow for all diodes (including LEDs) on this op-amp. For diodes with exceptionally high Rz, measurements should be conducted within a few minutes per diode reading.
The other end of the optic cable will go to another small shield that contains a photodiode, that goes to a circuit, that will go to an appropriate output connector where a voltage meter can connect to.
So, the new setup will once again have to main sections separated by a fiber optic cable.
Inside the Hammond shield will be ultra low power *latched* reed relays. Such relays require power for a very short duration, milli seconds, to toggle the latch. After that, no power is required to maintain that latch. The energy required to toggle the latch is ~ 500 uJ. The manufacturer datasheet shows these reed relays only produce 0.3 uV of thermal electric effect while powered. Since they are powered for a very short duration, there will be no measurable thermal electric effects. Furthermore, such infinitesimal thermal electric effects quickly fade since the relays require power momentarily. Also, the diodes of interest will produce DC voltages in the milli volts and most perhaps over one volt.
Such latched reed relays will be responsible for –>
The above is for the stage 4b setup where there will be just one diode (array) inside the Hammond. For stage 4a, there will only be two latched reed relays since there will be dozens of diode (arrays) inside the Hammond shield, which will require dozens of thin strings. Each string will connect a particular diode (array) to the electrometer by means of a copper contact switch.
Edit: This blog post is continued at New diode testing design [2].
My goal is to complete the stage 4b diode testing phase by the end of November, 2009. Hopefully within two to three weeks stages 4a and 4b will begin. Stage 4a will consist of testing dozens of unique diodes. Stage 4b is about taking a gamble that a specific diode will meet my minimum requirements that the specific diode will produce sufficient DC voltage in addition to recovering within 4 weeks time given a typical disturbance. I have not yet set the parameters as to what defines a “mild disturbance.”
Hopefully stage 4b will take no longer than four to five months to test the specific diode to see how it performs over time, and to see how quickly it recovers. Each recovery period could take up to four weeks. If the diode takes longer than four weeks to recover, then the gamble failed, in which case I’ll have to find another diode, and thus stage 4b starts over.
IMO 4 to five months is the best case scenario, as such diode experiments are like watching grass grow.
If I can afford it, I might buy another Hammond shield and conduct two stage 4b experiments simultaneously. This is *not* an invintation for money donations, as I absolutetly refuse to accept money donations for the research. Although people have loaned me parts and equipment.
After looking at two major manufactures, it appears the metalized mylar bags is not such a good idea after all. The metal layer is sandwiched between two layers of insulation, so there’s no way of properly sealing the bag.
I’m certain the DMM will not need shielding if the following method is used –>
This is easy to test simply roaming around with the mobile diode testing setup to see if the DC voltage varies. Try pushing the limits by going near a wi-fi that is transmitting. You can download a huge file on the Internet to make the wi-fi transmit.
If it’s not shielding enough, then you can place the setup in a slightly larger shield, thus forming two layers of shield, cut a rectangular hole in the outer shield where the LCD display is, and replace that with a metal wire mesh screen instead. This will provide a closed metal shield while allowing you to view the LCD.
If that’s insufficient shielding, then quickly move out of there because your health is at risk from intense radiation exposure. 
Edit: This is not such a good idea after. Please read why.
A EE sent me an email with a great idea! Put the DMM or just everything inside an Aluminum metalized bag –>
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=SCP162-ND
These bag have a thin metal layer that makes the mage semitraspent. The above bag has a transparency of 40%, which is more than sufficient.
Edit: This is not such a good idea after. Please read why.
Here’s the hot setup. Outline –>
Electrometer, batteries, and diodes to be tested are inside metal chassis. Electrometer is always off except for while measuring diode DC voltages. Electrometer output goes to *very* thin twisted copper enamel coated wire, which goes through a very small chassis pin hole. Immediately outside of the chassis right where the twisted wire comes out of the pin hole is a high input resistance op-amp, a FET op-amp will work great. Also connected to the twisted wire immediately outside the chassis is a capacitor and also a shunt resistor, which will shunt any measurable external RF noise. The FET op-amp output goes to a LCD panel meter or DMM.
This type of setup will prevent all measurable external RF noise from entering the chassis. The wires that go to the DMM and the DMM itself will pickup a small amount of RF noise, which will travel down the wires that go to the FET op-amp, which is nothing but a short relative to the FET input resistance. The circuit would be a capacitor, say 1uF, a resistor, say 150 ohms, and a 10Tohm (10e+12 ohms), all in parallel to each other. So which path do you think the current will take. It surely will not take the path through the 10Tohm resistance, which is the path from the FET op-amp output and input stage. So the capacitor and 150 ohm resistance short any external noise that would normally enter the chassis.
The reason the FET op-amp, capacitor, and shunt resistor need to be as close as possible to the chassis is to eliminate as much external noise pickup as possible. Pick a FET op-amp that is very small.
Here’s a simple schematic –>
The switch that is between the DUT and INA116PA is a simplified symbol. Ideally you want a two way switch such that it reverses the DUT relative to the INA116PA.
The components next to the FET op-amp are a capacitor and resistor, believe it or not. You might need a magnifying glass. 
Further details will be posted in future blogs. Details such as the method of turning the electrometer on and off, and the type of switches. I have used Mercury tilt switches. For my next new setup, I may continue to use such tilt switches, but for the Grand Final setup that will be taken to demonstrate the diode array to Universities and such will most likely use ultra low power small latched reed relay switches that will be momentarily powered by an LED shining light through a small pinhole and picked up by a photodiode.
Building a simple inexpensive 1N4148WS diode array to prove for yourself that diodes produce a DC voltage:
It should not take that many 1N4148WS diodes in-series to produce 1 mV. By my calculations, it would take ~ just 17 of the 1N4148WS undisturbed diodes connected in-series to produce 1 mV. The price of the 1N4148WS diode at a quantity of 17 is just $0.04 (4 cents) each. I would predict that these diodes should recover in ~ one month. Don’t get me wrong, if you produce a relative amount of current in the diodes (relative to what the diodes produce, which is ~ 10 pA), then I have no idea how long the diodes will take to reach the undisturbed state. What is meant by “disturbing” the diodes is on the order of a few nano amps.
When your diodes arrive by mail, immediately solder them together in-series. *PLEASE* only *tap* the soldering iron on the diode pins just long enough to get a basic solder connection. No need to enter you diode array in a beauty contest or take them four wheel driving. So a very quick solder job is sufficient to hold the diode pins together in-series. Make sure the diodes are soldered in the same polarity/direction. Heat quickly disturbs diodes, which is why it is important for best results to keep the soldering iron time to a minimum on the diodes. Also, do *NOT* place the diodes on a meter. The manufacturer has already tested the diodes. The odds of the diode not working is almost non-existent. The only meter you want to place on these is an electrometer. Please read the following blog post for details –>
How to replicate the diode experiments
Todays IR Photodiode test might be the last for awhile. From Hawaii, Charles M. Brown mailed me a THz diode array chip to test, made by Virginia Diodes. It just arrived a few minutes ago. I have already dismantled part of the IR Photodiode setup, and hope to start building the new multi diode testing setup today. A lot of designs have passed over the table, but I’ll use the old faithful electrometer. Why mess with something that already works far better than it needs.
The new setup will consist of some homemade *very* simple metal contacts. Each contact will be connected to a waxed dental floss string. Each contact will be two pieces of separated thin metal, probably Aluminum. A string will be connected to one of the metal foils. The waxed dental floss string will go through a thin hole in the inner shield, and also through the outer shield to the outside world. The outside end of the floss string will be tied to a small weight. For open-circuit, the weight will be on a little shelf so as not to pull on the floss string. To close the circuit, the small weight will be gently lowered to close the contacts. I’ve used this method before, and there’s probably nothing better, but it takes time to make the contacts.
The new setup will house a lot of diode tests, simultaneously. Most of the diodes will be off the shelf diodes, including Charles THz diode chip. Radio Shack sells a lot of different diodes, mostly LEDs. I would like test as many of such LEDs as possible. Most of the tests will be new diodes. I’m expecting such new diodes to be highly disturbed. So we’ll see how long it takes to recover. Diodes with lower Rz such as the 1N914 will probably recover first. Also, I might include my green LED even though it’s most likely disturbed.
Also a capacitor only test will be included. Also a battery will be included so as to calibrate the electrometer during every test.
Edit: Note, this design is now out of date, and has been replaced with New diode testing design.
One nice diode testing methods uses mechanical tilt switches (do not use Mercury switches, as their resistance is too low), that offers no measurable effect on the diode and the measured DC voltage. Such switches are $1.89 each. Here is a schematic of such a setup –>
I would recommend using 1.0uF low leakage capacitor. The process begins by slightly tilting the enter setup (that includes all of the metal shields). This will connect the diode with the low leakage capacitor. Then allow the diode sufficient time to charge the capacitor. About 10 hours for a 1.0 uF capacitor. After 10 hours of charging time, slightly tilt the entire setup again so as to turn on the electrometer which connects the battery to the electrometer. Give the electrometer some time to settle down, about 10 minutes is fine. In ten minutes, slightly tilt the entire setup a little more to connect the electrometer input to the low leakage capacitor & diode.
Of course, before doing the above, you must have the electrometer adjusted. Do this by replacing the above diode with ~ 0.15 voltage source. You can use a 1.5 volt battery connected to two resistors in series, 10 Kohm and 1 Kohm. Connect the low leakage capacitor across the 1Kohm resistor. That will charge the capacitor to ~ 0.15 volts. Do repeated cold starts by turning on the electrometer (when its at room temperature), and do the above test except by replacing the diode with the battery voltage source. Adjust the electrometer so that it shows the correct voltage. Repeat numerous times until you get consistent repeatable results. Then replace the 1 Kohm resistor with a 100 ohm resistor, and repeat the calibration tests again just to be on the safe side. You should get consistent 0.015 volt measurements. Then remove the battery voltage source and put in the diode.
I would recommend the Ina116P (pdf file) op-amp, not theIna116PA. The Ina116P should be held in the air by its wires so as to minimum the input bias current.
Loading ...
Loading ...